cognitive-ai

Cognitive AI

While we don’t yet have strong artificial intelligence that can learn new tasks completely by itself, handle tasks it isn’t pre-programmed for, or have self-awareness and emotions, it appears cognitive computing may be moving in that direction.

The most well-known cognitive system is probably IBM Watson, which famously won Jeopardy in 2011 over the game’s best players in the world.

Cognitive systems understand unstructured data and natural language, can interact with humans in a natural way, and sift through enormous amounts of unstructured data, providing suggestions and decision support in specific fields of expertise.

According to IBM, Watson augments human abilities by scaling and accelerating human expertise. In effect, it helps share experience at scale.

Cognitive systems are defined by how they understand natural text, speech, and have vision using image recognition. They use deep machine learning to learn almost everything about any topic at scale, and they can reason and provide suggestions to help people in their decision-making.

Cognitive computing enables the democratization of expertise by taking the knowledge of an industry’s best experts and sharing it with others. In effect, these systems are decision support tools with skills in particular topics, say cancer treatment, tax auditing, or something else entirely.

They can help share the highest levels of expertise outside the core competence centers and help make this knowledge accessible to a wider audience.

For example, a cancer diagnostic decision system could be trained with the majority of the worlds combined knowledge on the matter. It can then be used by small local hospitals in rural areas or by doctors in poor countries on the other side of the world.

This could improve the quality and efficiency of advanced medical treatment at a grand scale.

For our purposes, IBM Watson also brings cognitive computing to marketers, offering capabilities such as customer journey analysis, real-time personalization, surfacing marketing insights, and cognitive content management.

While IBM pushes the built-in marketing capabilities of Watson, this is not the only interesting aspect of the platform. Perhaps more importantly, IBM allows developers to build their own domain-specific apps that use Watson as an AI-engine for cognitive computing.

This makes it relatively simple to design advanced cognitive AI systems by building applications on top of IBM Watson.

This is what Equals3 did to create Lucy, an AI agent that can perform market research and harvest insights, provide marketing decision support, and facilitate automatic generation of buyer personas and media strategies.

Other companies do the same, for example to create smart in-store digital signage or information kiosks that can adapt their content and tone of communication dependent on who uses them.

Perhaps we will see only a handful of advanced AI systems from the big vendors in the future, and all other AI tools will be built as apps on top of them. Time will tell, but it is certainly an interesting thought.

With that model, any small business could build super-smart cognitive AI systems for particular niches, using AI platform technology well beyond what they could ever develop from scratch by themselves.

Either way, cognitive AI might have enormous effects on the democratization of knowledge and expertise in the future, as everyone could have the world’s combined knowledge on virtually any topic accessible through a simple chat or voice interface.

Broadening access to these high-level tools will likely have wide-ranging implications as they develop.